Before you install

  • A Windows 10 or 11 x64 PC that already holds the project files.
  • The Word, Markdown, or LaTeX draft you will submit or revise.
  • Supporting data tables and the papers you cited, in the same folder if you have them.
  • A network connection for account activation, AI review, and online search.

Download the installer from this site and, if you need to verify it, compare the SHA-256 on the download page. macOS is not available yet.

Open the project and import the draft

After installation, choose the folder that already contains the paper. That folder is the project; you do not create a separate cloud library.

Import the draft you are revising so the check has a manuscript to follow. The original file remains read-only. Review takes place on a working copy in the app. If you later change the same Word file outside MediCrab, the app will not silently overwrite either side.

Run the check and read the list

Start the full-manuscript check from the Checks area. The list groups what is verified, what needs attention, and what is currently uncovered. Citation lookups, number mismatches, and consistency findings appear together so you are not hopping between tools.

Open a finding to see the evidence next to it—the manuscript span, the source record, or the reason the lookup stopped. Simulated homepage copy such as “Abstract n = 410 · Results n = 412” is an illustration of this view, not a real cohort.

Confirm changes; do not skip the evidence

Suggested edits do not rewrite your original file. You read the finding, check the source, and only then accept a change. Conclusion-level edits in particular should be confirmed one by one: a wrong accept is hard to notice on a later reread.

First-time account activation includes promotional compute credits. Model-assisted review and online literature lookup consume credits according to usage. The app does not guarantee how many full checks that balance will cover.

Anything sent to a model or search provider is shown and recorded. Omit patient identifiers and unpublished data your institution does not allow to leave the machine.